The Digital Farms of British Columbia

Data centres sized around communities.

Modular, liquid-cooled AI infrastructure with recovered heat, dry-cooler backup and a dedicated optical fabric between towns.

Thermal architecture

Direct liquid cooling

Warm-water liquid loops improve heat capture and reduce dependence on high-airflow fan systems.

Heat cascade

Highest-value heat first

Greenhouse heating, wash/process water and district loops take priority; dry coolers reject only what cannot be economically used.

Network

Optical scale-across

Dedicated fibre and wavelength services connect sites for distributed workloads, replication and regional scheduling.

A simple 10 MW heat example

A continuously loaded 10 MW IT block consumes 87.6 GWh of IT electricity per year. Nearly all of that electrical energy ultimately becomes heat. After allowing for capture, piping and end-use losses, a practical screening range of roughly 65–80 GWh/year of useful low-temperature heat is reasonable for early planning. Final numbers depend on coolant temperatures, annual load factor, heat-user return temperatures and distance.

This is an energy-balance screening example, not a guarantee of saleable heat. Summer heat demand can be much lower than winter demand, so storage and alternate heat sinks matter.

Cooling hierarchy

  1. Serve contracted heat users.
  2. Charge hot-water or seasonal thermal storage where economic.
  3. Use economized dry cooling in cold weather.
  4. Use efficient hybrid/evaporative assist only when weather and water availability justify it.
Why smaller sites can win

Permitting and heat demand become design inputs.

Distributed sites trade some campus-scale economies for better access to existing heat users, smaller interconnection steps and phased capital deployment.

Rural data centre conceptData centre heat reuse conceptFibre corridor concept